Skip to main content
Download PDF

First-in-human, double-blind, randomized phase 1b study of peptide immunotherapy IMCY-0098 in new-onset type 1 diabetes: an exploratory analysis of immune biomarkers

BMC Medicine volume 22, Article number: 259 (2024) Cite this article

Abstract

Background

IMCY-0098, a synthetic peptide developed to halt disease progression via elimination of key immune cells in the autoimmune cascade, has shown a promising safety profile for the treatment of type 1 diabetes (T1D) in a recent phase 1b trial. This exploratory analysis of data from that trial aimed to identify the patient biomarkers at baseline associated with a positive response to treatment and examined the associations between immune response parameters and clinical efficacy endpoints (as surrogates for mechanism of action endpoints) using an artificial intelligence-based approach of unsupervised explainable machine learning.

Methods

We conducted an exploratory analysis of data from a phase 1b, dose-escalation, randomized, placebo-controlled study of IMCY-0098 in patients with recent-onset T1D. Here, a panel of markers of T cell activation, memory T cells, and effector T cell response were analyzed via descriptive statistics. Artificial intelligence-based analyses of associations between all variables, including immune responses and clinical responses, were performed using the Knowledge Extraction and Management (KEM®) v 3.6.2 analytical platform.

Results

The relationship between all available patient data was investigated using unsupervised machine learning implemented in the KEM® environment. Of 15 associations found for the dose C group (450 μg subcutaneously followed by 3 × 225 μg subcutaneously), seven involved human leukocyte antigen (HLA) type, all of which identified improvement/absence of worsening of disease parameters in DR4+ patients and worsening/absence of improvement in DR4− patients. This association with DR4+ and non-DR3 was confirmed using the endpoints normalized area under the curve C-peptide from mixed meal tolerance tests where presence of DR4 HLA haplotype was associated with an improvement in both endpoints. Exploratory immune analysis showed that IMCY-0098 dose B (150 μg subcutaneously followed by 3 × 75 μg subcutaneously) and dose C led to an increase in presumed/potentially protective antigen-specific cytolytic CD4+ T cells and a decrease in pathogenic CD8+ T cells, consistent with the expected mechanism of action of IMCY-0098. The analysis identified significant associations between immune and clinical responses to IMCY-0098.

Conclusions

Promising preliminary efficacy results support the design of a phase 2 study of IMCY-0098 in patients with recent-onset T1D.

Trial registration

ClinicalTrials.gov NCT03272269; EudraCT: 2016–003514-27.

Peer Review reports

Background

Type 1 diabetes (T1D) is a chronic autoimmune disease that accounts for approximately 5–10% of all diabetes diagnoses and > 85% of the diagnoses made in youth [1, 2]. The disease is characterized by loss of sensitivity to insulin-producing β cells, which leads to β cell destruction, a decline in endogenous insulin secretion and, consequently, hyperglycemia [3]. Therapies that aim to modify the underlying cause of disease, such as antigen or antibody immunotherapies, are of particular interest and have demonstrated efficacy and safety in previous studies [4,5,6]. IMCY-0098 is an Imotope™, a linear synthetic peptide encompassing a proinsulin C20-A1 epitope and a thioredox motif, which has been shown to induce a cytolytic phenotype in a human CD4 T cell line and is able to eliminate antigen-presenting cells (APCs) that present the proinsulin epitope (data not shown). Interestingly, treatment with a mouse insulin-derived Imotope™ offered protection from diabetes in a non-obese diabetes mouse model [7]. Furthermore, cytolytic CD4 cells also eliminate autoreactive pathogenic T cells upon cognate interaction with the same APCs, irrespective of the epitope presented [7, 8]. The IMCY-0098 peptide has shown good binding capacity to DR3 and DR4 polymorphisms, which are strongly associated with T1D [9, 10].

Machine learning is an important component of artificial intelligence that has been integrated into many fields, including drug discovery, drug development, and the prediction of drug efficacy [11]. The Knowledge Extraction and Management (KEM®) analytical platform is a machine learning system used to systematically extract unsupervised relationships between collected variables with no predefined hypothesis [12]. KEM® artificial intelligence uses the formal concept analysis framework [13] that has been successfully applied in different domains, including drug discovery, studies for the identification of patient selection biomarkers for therapeutic responses and genomic characterization of complex diseases [14,15,16,17,18]. KEM® systematically identifies all groups of patients with shared characteristics and generates corresponding association rules that can detect patterns and relationships in heterogeneous databases [18,19,20,21,22]. This approach (formal concept analysis) is different from predictive modeling as it allows for identification of the most relevant hypothesis that is consistent with the data [23]. Formal concept analyses are able to extrapolate significant relational detail from datasets with small sample size and a large number of variables [24], and association rules help to detect rare signals. Drawing conclusions from small sample sizes is of importance for medical research [25], and these methods may be useful for this purpose.

Data from a recent phase 1b trial (NCT03272269) demonstrated a promising safety profile of ICMY-0098 [10]. In this exploratory analysis of samples from this trial, we aimed to use an artificial intelligence-based approach, suitable for small patient numbers, of unsupervised explainable machine learning to explore the associations between all variables. Specifically, we aimed to identify the biomarkers at baseline associated with positive response to treatment and to examine the associations between immune response parameters and clinical efficacy endpoints as surrogate endpoints for mechanism of action. This study aimed to identify patient and disease characteristics associated with response to treatment with IMCY-0098 using the KEM® analytical platform, to gain further insights into the mode of action of IMCY-0098, and to inform the design of future clinical studies.

Methods

Study design

This was an exploratory analysis of data from a phase 1b, dose-escalation, randomized, placebo-controlled study of IMCY-0098 in patients with recent-onset T1D (NCT03272269). The study design was as described previously [10]. Briefly, 41 patients diagnosed with T1D ≤ 6 months before the start of the study were randomized 3:1 to receive IMCY-0098 or placebo. Patients allocated to receive IMCY-0098 were sequentially enrolled to receive dose A (50 μg subcutaneously followed by 3 × 25 μg subcutaneously), dose B (150 μg subcutaneously followed by 3 × 75 μg subcutaneously), or dose C (450 μg followed by 3 × 225 μg subcutaneously). Treatment was administered in four doses with aluminum hydroxide adjuvant (alum) at a concentration of 500 μg/mL in 2-week intervals from weeks 0 to 6 with follow-up at weeks 12, 18, and 24. The study included patients who were HLA DR3 + and/or DR4 + ; had ≥ 1 autoantibody against GAD65, IA-2, or ZnT8; and had fasting C-peptide at screening (week −4) > 0.2 nmol/L and/or stimulated C-peptide ≥ 0.4 nmol/L. Data were collected in a double-blind manner until all patients completed the study to week 24 or were prematurely withdrawn from the study.

Study objectives and endpoints

The primary objective of the phase 1b study was to assess the safety of IMCY-0098 and the secondary objective was to assess the clinical response (previously published [10]). There was an additional exploratory objective to characterize the immune response to IMCY-0098 treatment using a panel of markers of T cell activation, memory T cells, and effector T cell response. As such, the immune endpoints described in the current study included detection of CD4+ T cells specific for epitope C20-A1 of proinsulin (peptide sequence included in IMCY-0098), impact on effector CD4+, and CD8+ T cell responses specific for insulin, glutamic acid decarboxylase 65 (GAD65), and islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP).

The main time point for all endpoints was 24 weeks; however, other time points were also considered in the analysis.

Study procedures and assessments

Exploratory immunologic assessments were performed in a specialized and independent laboratory. Approximately 100 mL of blood was collected at each time point (weeks −4, 0, 6, 12, and 24) for peripheral blood mononuclear cells (PBMC) preparation. All immunoassay analyses were performed using fluorescence-activated cell sorting techniques.

Clinical response

A clinical response parameter, expected area under the curve (AUC) C-peptide, was derived based on the difference between the normalized AUC measured during mixed meal tolerance tests (MMTT tests) and the AUC expected values given the general disease evolution, as described previously [26]. The AUC expected values used in this study were chosen based on the time to diagnosis (< 12 months for all 41 patients): −0.0245 pmol/mL/month (95% confidence interval: −0.0271, −0.0215).

Immune response

PBMCs were isolated from blood using Ficoll density gradient centrifugation and analyzed within 24 h of blood collection or after in vitro stimulation that aimed to expand specific T cells with a cocktail of all peptides (Peptivator® human insulin, Miltenyi, Bergisch Gladbach, Germany), GAD65 (Peptivator® human GAD65, Miltenyi, Bergisch Gladbach, Germany), three IGRP major epitopes (Lifetein, Somerset, NJ, USA), or the C20-A1 epitope of proinsulin included in IMCY-0098. Negative and positive controls were of peptide diluent or Staphylococcal enterotoxin B, respectively. Freshly isolated or in vitro expanded PBMCs were stimulated with indicated antigens (using a 12-day expansion phase with restimulation 18 h before staining) for subtype markers (CD3, CD4, CD8), naïve/memory T cell differentiation markers (CCR7, CD45RA), and proliferation marker (antigen Kiel 67 [Ki67]) and function (interleukin [IL]-17, interferon gamma [IFN-γ], lysosomal-associated membrane protein-1 [CD107a], granzyme B and perforin), before flow cytometry analysis on a BD® LSR cytometer (Becton Dickinson, Erembodegem, Belgium). A summary of the assay methodology is presented in Fig. 1.

Fig. 1
figure 1

Summary of immune response methodology

Full size image

For detection of specific CD8+ and CD4+ T cells after the in vitro expansion phase, HLA-A*02 pentamers loaded with Preproinsulin, IGRP, or GAD65 epitopes; DRB1*03:01 tetramer loaded with PI C20-A1 peptide or GAD65 peptide; and DRB1*04:01 tetramer loaded with K to S substitution in position 9 of PI C20-A1 peptide [27, 28] or GAD65 peptide were combined with multiparameter flow cytometry. T cell subsets were analyzed using specific markers: naïve (CD45RA + CCR7 +), central memory (CD45RA-CCR7 +), effector memory (CD45RA-CCR7–), and terminal effector (CD45RA + CCR7–).

Delta ratios

For both clinical and immune endpoints that were measured at different timepoints, delta ratios were calculated. Firstly, delta ratios were calculated using the following formula to measure the evolution of the disease from baseline, where Vn = visit and Vbaseline = baseline visit:

$$\Delta{endpoint} \left(V{baseline} \to Vn\right)=\frac{\,{endpoint} \left(Vn\right)- {endpoint}\,\left(V{baseline}\right)}{{endpoint} \,(V{baseline})}$$

Vn = visit n; Vbaseline = baseline visit (randomization).

Secondly, all numerical parameters (delta ratios and baseline characteristics) were discretized as described here: (1) delta ratios were discretized in three categories (increasing [delta > 0], stable [delta = 0], or decreasing [delta < 0]), and (2) baseline characteristics were categorized using an equal-frequency tertile binning as high (top tertile), medium (middle tertile), and low (bottom tertile).

Statistical analysis

The study sample size for this phase 1 study is not based on any statistical hypothesis testing, as is often the case for first-in-human studies where there is little information available beforehand. However, it was aligned with a similar study of peptide immunotherapy [29] and determined as adequate to support preliminary assessment of safety, efficacy, and immune response. This size favored a higher patient number in the dose C group as this higher dose was expected to achieve greater immunogenicity versus a lower-dose group, an important consideration for the current analysis [10]. This approach was fully endorsed by seven regulatory authorities.

Formal concept analysis-based artificial intelligence

Artificial intelligence-based association rule analyses between immune responses and clinical responses were exploratory in nature; the analysis was included in the study protocol/statistical analysis plan, but no detailed method was pre-specified.

The KEM® v 3.6.2 analytical platform is an explainable AI platform that systematically implements unsupervised mining of association rules and was used to systematically extract unsupervised relations between all variables collected with no predefined hypothesis [12, 19, 30]. An association rule is defined as a relationship X → Y, where X (left part) and Y (right part) can be a unique descriptor or combination of descriptors (Fig. 2). Five quality measures were used to characterize and rank all generated rules: support (the number of patients for whom the association was identified), confidence (the probability of having Y within patients verifying X), lift (the ratio between confidence and the probability of having Y in the whole dataset; lift describes the enrichment brought by considering patients verifying X instead of all patients), and two p-values (derived from Fisher’s exact test and Mann–Whitney–Wilcoxon). Statistical significance was neither necessary nor sufficient to indicate a causal relationship. Multiple tests were not taken into account in the rule-of-association approach; formal concept analysis methods generate relationships between descriptors in an exhaustive manner despite large combinatorial space and compute metric values for each relationship, so multiple analyses/corrections are not necessary.

Fig. 2
figure 2

Example application of quality measures support, confidence, and lift to association rules

Full size image

Association between clinical outcomes at visit 8 (week 24), dose, and the evolution of immune markers was explored by focusing on the subset of relations where dose was necessary and sufficient. In addition, relations between all available descriptors were systematically characterized, and the associations between treatment groups, clinical outcomes (combination of antecedents), and immune response (consequent) were analyzed using the KEM® platform (Additional File 1: Fig. S1).

Post hoc analyses were performed to assess the association with DR4+ and non-DR3 using the following endpoints: normalized AUC C-peptide from MMTT, fasting C-peptide/glucose and insulin. Given the small study sample size and the HLA imbalance between groups, these analyses were performed by grouping the placebo group with dose A (untreated group) and dose B with dose C (treated group). An additional responder analysis was performed using the published quantitative response (QR) methodology adjusted based on data from the 6-month follow-up [31]. For the normalized AUC C-peptide, this model allows the calculation of an expected C-peptide value after a period based on baseline AUC and age at T1D onset. The observed value minus the model-based expected C-peptide value (QR) was defined to reflect the effect of the therapy.

Results

Patient HLA haplotypes are presented in Additional File 1: Table S1. The most common HLA type was DR4+ only (46.3%), and most patients were HLA-A2:01 (63.4%).

Clinical response

KEM® was used to systematically generate all association rules between the descriptors of the dataset (treatment dose and candidate subgroups, defined based on patient characteristics) and the endpoints (clinical or immune response at any visit) (Fig. 3A).

Fig. 3
figure 3

Formal concept analysis of associations between treatment group, clinical endpoints, immune response data, and subgroups. A Analysis workflow. Among all 32,079 generated rules, 12,638 relevant rules were explored to find associations for clinical response. Associations were selected for support ≥ 4, lift ≥ 1.25, confidence ≥ 0.75, and p ≤ 0.05. B Statistically significant associations involving treatment with IMCY-0098 dose C, improvement of clinical endpoints, and HLA type. C, D HLA-dependent changes in subjects among the dose B and dose C groups for C normalized AUC C-peptide from MMTT and fasting C-peptide/glucose and D insulin dose per kg. E Box plot of quantitative response for AUC C-peptide in untreated vs treated groups. Central lines represent median values, boxes represent interquartile range, and whiskers represent upper and lower 1.5 × interquartile range, respectively. Dose B: 150 μg at week 0 followed by 3 × 75 μg; dose C: 450 μg at week 0 followed by 3 × 225 μg. * Fischer’s p = 0.044. AUC, area under the curve; HLA, human leukocyte antigen; MMTT, mixed meal tolerance test

Full size image

The analysis was focused on two endpoints selected for general knowledge, signal strength, and interpretability: fasting C-peptide/glucose ratio and the difference between the measured and the expected normalized AUC for C-peptide from MMTT. After applying the filters, 42 associations involving 18 clinical markers were selected (Additional File 1: Table S2). Of 15 associations found for the IMCY-0098 dose C treatment group, seven involved HLA type. All of the associations identified improvement/absence of worsening of disease parameters in DR4+ or DR3− patients (two associations at two different visits) and worsening/absence of improvement in DR4− patients (two associations at Visit 6) (Fig. 3B).

The proportion of responders among the treated group increases based on HLA DR4 positivity and absence of DR3. When focusing on the evolution of normalized AUC C-peptide from MMTT in DR3− patients, this proportion is significantly higher compared with the untreated group: Fisher’s p = 0.044 (Fig. 3C). The grouping used for the analyses—the treated (i.e., dose B + C) and untreated (i.e., placebo + dose A) groups—was supported by the absence of any association rules regarding the HLA status of patients in the placebo and dose A groups, while the dose B and C groups shared common associations from the KEM analysis.

The postulated association with DR4 + and non-DR3 was confirmed using the insulin endpoint; the presence of DR4 HLA type and the absence of DR3 HLA type were associated with improvement of this endpoint in the post hoc analysis (Fig. 3D). In another post hoc analysis we looked at the normalized AUC C-peptide using the QR model approach. Briefly, this model allows calculation of an expected C-peptide value after a period based on baseline AUC and age at T1D onset. The observed minus the model-based expected C-peptide value (QR) is defined to reflect the effect of the therapy. The results showed that the quantitative response in the treated group was significantly higher versus the untreated group (Fig. 3E).

Immune component analysis

Immune cells were analyzed using multiparameter multicolor flow cytometry for markers of T cell subtype, activation, proliferation, and function (Additional File 1: Table S3). Immune parameters following IMCY-0098 treatment and placebo were compared using three readouts: prevalence of general T cell subsets (to determine any unspecific treatment effect), induction of treatment (IMCY-0098)-specific cytolytic CD4+ T cells, and any changes in pathogenic T cells specific for β cell autoantigens. Across treatment groups, minor variations in the frequency of different T cell populations were observed during the study period. The numbers were similar for patients receiving IMCY-0098 and those receiving placebo, indicating that treatment with IMCY-0098 did not promote non-specific expansion or deletion of T cell subsets (Fig. 4A). These data support that the mode of action of IMCY-0098 is antigen-specific and targeted.

Fig. 4
figure 4

Immune response to IMCY-0098 treatment—intent-to-treat population. A Frequencies of different T cell subsets over time: naïve (CD45RA + CCR7 +), central memory (CD45RA-CCR7 +), effector memory (CD45RA-CCR7-), and terminal effector (CD45RA + CCR7 − ; no in vitro stimulation; error bars represent SD). B Change from baseline to week 24 in treatment-specific granzyme B+ cells within CD4 + T cell subsets after in vitro stimulation with IMCY0163 (insulin C20-A1). C Distribution of CD4+ granzyme B + T cells at week 24. D Change from baseline to week 24 in disease-specific Perforin + CD8+ T cells after in vitro stimulation with GAD65 and IGRP peptides. aFrequency within total CD4+ cell population. bFrequency within total CD8+ cell population. Box and whisker plots (B, D) show change of cell numbers from baseline to week 24, normalized to baseline value. Central lines represent median values, boxes represent interquartile range, and whiskers represent upper and lower 1.5 × interquartile range, respectively. p-values were obtained using Wilcoxon-Mann–Whitney test; p > 0.05 unless indicated otherwise. Dose A: 50 μg at week 0 followed by 3 × 25 μg; dose B: 150 μg at week 0 followed by 3 × 75 μg; dose C: 450 μg at week 0 followed by 3 × 225 μg. SD, standard deviation

Full size image

The change from baseline to week 24 in the number of granzyme B+ CD4+ cytolytic T cells in response to in vitro stimulation with the proinsulin epitope C20-A1 showed an upward trend in samples from IMCY-0098-treated groups (dose B and C) compared with placebo (Fig. 4B). A statistically significant difference was observed for IMCY-0098 dose B and dose C vs placebo for the change in CD4+ terminal effector cells (p < 0.001 and p = 0.047, respectively); differences did not reach statistical significance for central memory or effector memory CD4+ cells or for CD4+ cells overall (Fig. 4B). Importantly, central memory and effector memory cells were present across treatment groups at week 24 included, indicating that the cytolytic CD4+ T cell response may be sustained long term (Fig. 4C).

The number of Perforin+ CD8+ pathogenic T cells specific for β cell autoantigens measured following in vitro stimulation with GAD65 and IGRP showed varying trends with different doses of IMCY-0098 treatment (Fig. 4D).

Immune response: unsupervised formal concept analysis

Several statistically significant associations were identified between immune response parameters and IMCY-0098 treatment (Fig. 5A, Fig. S2). Treatment with IMCY-0098 dose B and C showed a positive association with increases in unstimulated granzyme B+ CD4+ T cells as well as in granzyme B+ CD4+ and IL-17+ CD4+ T cells that were treatment-specific (responding to in vitro stimulation with proinsulin epitope C20-A1 that is contained in the IMCY-0098 sequence). Treatment with IMCY-0098 dose B showed an increase in Perforin+ CD8+ pathogenic T cells that responded to in vitro stimulation with GAD65/IGRP peptides, whereas dose C showed a reduction (Fig. 5B and C); this latter association met the significance criteria of the algorithm. Taken together, these results suggest that IMCY-0098 dose C modulates immune response in T1D by promoting the generation of treatment-specific cytolytic CD4+ T cells and inhibiting pathogenic autoreactive CD8+ T cells.

Fig. 5
figure 5

Immune response to IMCY-0098 treatment. Summary of immune parameters identified during formal concept analysis as associated with treatment (A) and changes in treatment-specific granzyme B+ CD4+ T cells (B) and disease-specific Perforin+ CD8+ T cells (C) after in vitro stimulation in all patients versus DR4 subgroup (see also Fig. 3 and Additional File 1: Fig. S2). Box and whisker plots (B, C) show change of cell numbers from baseline to week 24, normalized to baseline values. Central lines represent median values, boxes represent interquartile range, and whiskers represent upper and lower 1.5 × interquartile range, respectively. p-values were obtained using Wilcoxon-Mann–Whitney test; p > 0.05 unless indicated otherwise. Disease-specific CD8+ T cells refers to all T cells specific to any the disease-peptide loaded multimers. Dose A: 50 μg at week 0 followed by 3 × 25 μg; dose B: 150 μg at week 0 followed by 3 × 75 μg; dose C: 450 μg at week 0 followed by 3 × 225 μg

Full size image

The change from baseline to week 24 in granzyme B+ CD4+ cytolytic T cells expanded in vitro using proinsulin epitope C20-A1 was significantly higher in DR4+ patients treated with IMCY-0098 dose B or dose C vs placebo (p = 0.032 and p = 0.013, respectively); statistical significance was not reached for the overall patient population (Fig. 5B). The change from baseline to week 24 in Perforin+ CD8+ pathogenic T cells expanded in vitro using GAD65/IGRP peptides was significantly higher in patients treated with IMCY-0098 dose B vs placebo (p = 0.006 and p = 0.01 for all patients and for DR4+ subgroup, respectively; Fig. 5C). This change was significantly lower in patients treated with IMCY-0098 dose C vs placebo (p = 0.031 and p = 0.028 for all patients and for DR4+ subgroup, respectively; Fig. 5C). Reduction of pathogenic CD8+ T cells in response to IMCY-0098 dose C treatment was noted for several other HLA type subgroups: DR3−, DR3/DR4, DR4X (data not shown).

Association between clinical and immune response identified using formal concept analysis

Improvement and worsening of several clinical endpoints showed significant associations with increases and decreases of immune cell markers in the dose B and C groups (Fig. 6; Fig. S3). For example, in treatment groups IMCY-0098 dose B and dose C, decrease in insulin use showed a positive association with the increase in granzyme B+ CD4+ cytolytic T cells after in vitro expansion using insulin C20-A1 epitope (the natural epitope included in IMCY-0098: Fig. S4). In addition, increase of daily insulin dose was significantly associated with an increase in CD8+ pathogenic T cells after in vitro expansion with peptides derived from insulin, GAD65, or IGRP (data not shown). Similar trends were observed for associations between increase in various subpopulations of CD8+ pathogenic T cells and worsening of other clinical endpoints, such as fasting C-peptide and glucose levels and frequency of hypoglycemic events (data not shown). Conversely, in the placebo group, only very few non-relevant associations could be identified (Fig. S3).

Fig. 6
figure 6

Association between clinical outcomes and immune response markers. Results are shown for week 24 for patients receiving IMCY-0098 dose B or C. Associations were selected for confidence ≥ 0.75, support ≥ 4, and p ≤ 0.05. Data for IMCY-0098 dose B and dose C treatment groups were pooled for this analysis. Dose B: 150 μg at week 0 followed by 3 × 75 μg; dose C: 450 μg at week 0 followed by 3 × 225 μg. GAD, glutamic acid decarboxylase; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein; IFN, interferon

Full size image

Together, these results suggest that clinical improvement in patients treated with IMCY-0098 was associated with an increase in protective immune response (cytolytic T cells) and a reduction in autoimmune response (pathogenic T cells) (Fig. 6 and Additional File 1: Fig. S3).

Discussion

A growing body of literature suggests the existence of different endotypes and data-driven analyses help identify responders to treatment in diabetes [26, 32,33,34,35]. Previous studies applying multivariable analysis to clinical outcomes in T1D showed that certain baseline characteristics of patients and their disease may be associated with long-term outcomes [26]. To overcome the limitation of small patient numbers and to obtain further insights into patient outcomes with IMCY-0098 treatment, our exploratory analysis included artificial intelligence-driven analysis of clinical and immune response using KEM® [12, 19, 30, 36]. Although not specifically designed for the analysis of clinical trial data, rule extraction algorithms are common tools widely used in data mining in various domains. This returned a number of statistically significant associations between treatment and clinical outcomes and between immune response and clinical outcomes. In particular, we found multiple statistically significant associations between treatment with IMCY-0098, generation of potentially protective cytolytic CD4+ T cells, and inhibition of pathogenic autoreactive CD8+ T cells, suggesting that IMCY-0098 may elicit protective immune responses in humans.

Doses B and C showed similar increases in cytolytic CD4+ T cells, but only dose C significantly downregulated autoreactive CD8+ T cells. There are many possible reasons for this difference; phenotypic traits of cytolytic CD4+ T cells aside, lytic markers (such as granzyme B and CD107a) could differ slightly and thereby influence efficacy. Dose B was associated with an expansion of more terminally differentiated CD4+ T cells which suggests decreased effector capacity and weaker potency with which to control the immune response. This dosing effect should be evaluated in further studies with increased dosages and deeper phenotypic analysis of induced cytolytic CD4+ T cells by single cell analysis or cytometry by time of flight (CyTOF) technologies. It is also important to note that the impacted CD8+ T cells were of other specificities than treatment-induced CD4+ T cells, suggesting the action of a potent bystander killing mechanism.

Previous studies of genetic predisposition and recent studies assessing the benefit of immunotherapy for T1D suggest that patients’ HLA type could play a role in clinical efficacy [29, 35, 37]. A recent report indicated that DR3+ and DR4+ individuals may have a different course of disease, reflected, for example, by the difference in their response to teplizumab treatment [33]. Another recent study found that HLA type significantly influenced the effect of GAD65-alum therapy, with the best response to therapy observed in DR3+ DR4− individuals [35].

Patients enrolled in the current study were not stratified by their HLA haplotype, and the treatment groups were not balanced for individuals positive for DR3 only, DR4 only, or DR3/DR4. Despite this limitation, the results of the subgroup analysis for these HLA types showed that patients who did not express HLA DR3 and who received IMCY-0098 doses B and C tended to have positive clinical outcomes. Favorable tendencies were also observed for patients who did express HLA DR4; however, owing to the study design, these two subgroups were nested: HLA DR3− patients were all positive for DR4. To assess the difference between DR4-positive and DR3-negative patients, the analysis further explored the outcomes for DR3DR4 patients (DR4+ and DR3+): these patients did not show the favorable evolution that was observed on other DR4+ patients. Yet, given the small number of patients in this group (n = 5 for dose B + C), this difference between DR4+ and DR4+DR3− patients must be interpreted with caution. As such, no firm conclusion regarding the differential impact of DR3 and DR4 on IMCY-0098 efficacy could be drawn. The difference in clinical outcomes between DR3 and DR4 subgroups may be due to disease heterogeneity [38]. Furthermore, the lack of clinical benefit observed for DR3 in this study may be due to the timeframe over which the study was conducted as the disease may progress slower in the DR3 (compared with the DR4) subgroup [39]. Another potential explanation may be observed in the immune response to treatment; both subgroups increased treatment-induced cytolytic CD4+ T cells, but the DR3 subgroup may have had a smaller increase than the DR4 subgroup, which was not considered in the KEM® analysis. Finally, although in vitro characterization of IMCY-0098 indicated that it could bind similarly to DR3 and DR4 molecules [10], analysis of peptide-major histocompatibility complex (pMHC) stability showed that DR3 pMHC was less stable than DR4 pMHC (Imcyse data on file; data not shown). As pMHC stability influences the induction of the immune response [40], this may have in turn influenced the size of the cytolytic CD4+ T cell population in the DR3 subgroup.

In this exploratory analysis, clinical outcomes correlated with immune response in the IMCY-0098 dose B and C treatment groups, whereby improved clinical parameters were associated with an increase in cytolytic granzyme B+ CD4+ T cells and a decrease in pathogenic Perforin+ CD8+ T cells. Importantly, the increase in cytolytic CD4+ T cells appeared to be protective and did not deteriorate signs of disease. Recently, it has been reported that cytotoxic CD4 T cells are associated with T1D progression [41], but this was not observed in the current study. It is currently unknown if these cytotoxic CD4 T cells [41] are similar to the cytolytic CD4 T cells that increase upon treatment with IMCY-0098. Furthermore, although β cells have been reported to express MHC class II, there is no convincing evidence that they take up, process, and present auto-antigen-derived peptides [42]. If β cell MHC class II presentation was a predominant feature in T1D patients, we would have expected to observe disease exacerbation with IMCY-0098 treatment, as the MHC class II epitope IMCY-0098 contains (i.e., C20-A1) is known to be naturally processed from proinsulin [43]. Therefore, it is highly unlikely that cytolytic CD4+ T cells would kill β cells. These results support the probability that, at high doses, IMCY-0098 induces proinsulin-specific protective cytolytic CD4+ T cells and the subsequent elimination of pathogenic cytotoxic CD8+ T cells. This is consistent with the expected mechanism of action and the previous observations made in animal models [7].

The clinical data from the current trial have been published previously and showed that treatment with IMCY-0098 induced a reduction in some antibodies against T1D auto-antigens, suggesting a possible impact on specific T follicular helper cells by bystander killing [10]. The KEM® platform analysis showed an IMCY-0098 dose-specific association with CD4 cells expressing IFN-γ and/or IL-17 (Additional File 1: Fig. S2). As Th17 cells are generally associated with autoimmune processes, the observed association could look counterintuitive; however, it has recently been described that different subtypes of Th17 cells exist [44]. This highlights the need for more precise characterization of major T cell populations involved in the T1D pathology process following treatment with IMCY-0098, such as single cell approaches that can be implemented in a longitudinal study follow-up.

The exploratory nature of this analysis and small sample size were limitations of this study. Increases in treatment-specific FOXP3+ Treg cells were not expected based on the postulated mechanism of action; therefore, the current study did not examine these cells. Further studies should examine these cells to evaluate the involvement of regulatory T cells and, to a broader extent, the different T-helper subtypes that could be impacted by IMCY-0098 treatment. Future studies should also test various treatment doses and schedules to demonstrate clinical proof-of-concept based on C-peptide secretion maintenance. While the results from a larger phase 2 trial are needed to support the findings presented here, the data-driven machine learning approach identified a number of hypotheses consistent with the clinical data, which will be incorporated into the experimental design of subsequent clinical studies. The preliminary results from this exploratory study are important to inform the design of further clinical studies of IMCY-0098 which will use selection of (or stratification for) HLA type and a more robust assessment of immune response focusing on the induction of cytolytic CD4+ T cells.

Conclusions

This analysis showed that IMCY-0098 induces antigen-specific cytolytic CD4+ T cells and reduces the numbers of pathogenic CD8 + T cells in patients with T1D receiving the highest treatment dose. Due to the small study size and the exploratory nature of the machine learning-based analysis, this finding should be confirmed in larger studies, which will also help establish whether IMCY-0098 treatment efficacy is influenced by the patient’s HLA type. Furthermore, our data suggest that IMCY-0098 is a targeted, antigen-specific therapy; this mode of action may provide a better safety profile than non-specific immunosuppression although this deserves further study.

Availability of data and materials

De-identified patient data can be provided to independent qualified researchers upon submission of a written application/research proposal that should be approved by the study sponsor.

Abbreviations

APC:

Antigen-presenting cell

AUC:

Area under the curve

GAD65:

Glutamic acid decarboxylase

HLA :

Human leukocyte antigen

IFN:

Interferon

IGRP:

Islet-specific glucose-6-phosphatase catalytic subunit-related protein

IL:

Interleukin

KEM® :

Knowledge Extraction and Management

MMTT:

Mixed meal tolerance test

PBMC:

Peripheral blood mononuclear cells

T1D:

Type 1 diabetes

References

  1. Global report on diabetes. World Health Organization: World Health Organization; 2016. Available from: https://www.who.int/diabetes/global-report/en/.

  2. Classification and Diagnosis of Diabetes. Diabetes Care. 2016;39(Supplement 1):S13–22.

    Google Scholar 

  3. Waldron-Lynch F, Herold KC. Immunomodulatory therapy to preserve pancreatic beta-cell function in type 1 diabetes. Nat Rev Drug Discov. 2011;10(6):439–52.

    Article  CAS  PubMed  Google Scholar 

  4. Nowak C, Hannelius U, Ludvigsson J. Association between treatment effect on C-peptide preservation and HbA1c in meta-analysis of glutamic acid decarboxylase (GAD)-alum immunotherapy in recent-onset type 1 diabetes. Diabetes, obes metab. 2022;24(8):1647–55.

    Article  CAS  PubMed  Google Scholar 

  5. LeFevre JD, Cyriac SL, Tokmic A, Pitlick JM. Anti-CD3 monoclonal antibodies for the prevention and treatment of type 1 diabetes: a literature review. AJHP. 2022. https://doi.org/10.1093/ajhp/zxac244.

  6. Ludvigsson J, Hjorth M, Chéramy M, Axelsson S, Pihl M, Forsander G, et al. Extended evaluation of the safety and efficacy of GAD treatment of children and adolescents with recent-onset type 1 diabetes: a randomised controlled trial. Diabetologia. 2011;54(3):634–40.

    Article  CAS  PubMed  Google Scholar 

  7. Malek Abrahimians E, Vander Elst L, Carlier VA, Saint-Remy JM. Thioreductase-containing epitopes inhibit the development of type 1 diabetes in the NOD mouse model. Front Immunol. 2016;7:67.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Malek Abrahimians E, Carlier VA, Vander Elst L, Saint-Remy JM. MHC class II-restricted epitopes containing an oxidoreductase activity prompt CD4(+) T cells with apoptosis-inducing properties. Front Immunol. 2015;6:449.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Noble JA, Valdes AM. Genetics of the HLA region in the prediction of type 1 diabetes. Curr Diab Rep. 2011;11(6):533–42.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Van Rampelbergh J, Achenbach P, Leslie RD, Ali MA, Dayan C, Keymeulen B, et al. First-in-human, double-blind, randomized phase 1b study of peptide immunotherapy IMCY-0098 in new-onset type 1 diabetes. BMC Med. 2023;21(1):190.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Patel L, Shukla T, Huang X, Ussery DW, Wang S. Machine learning methods in drug discovery. Molecules. 2020;25(22):5277.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Jullian N, Jourdan N, Afshar M. Hypothesis Generation for Scientific Discovery. Examples from the use of KEM®, a rule-based method for multi-objective analysis and optimization. In: Kruse CG, Timmerman H, editors. Towards Drugs of the Future: Key Issues in Lead Finding and Lead Optimization. 9: IOS Press; 2008. p. 75–80. https://doi.org/10.3233/978-1-58603-949-3-75.

  13. Ganter B, Wille R. Formal concept analysis1999. https://doi.org/10.1007/978-3-642-59830-2.

  14. Abtroun L, Bunouf P, Gendreau RM, Vitton O. Is the efficacy of milnacipran in fibromyalgia predictable? A data-mining analysis of baseline and outcome variables. The Clin J Pain. 2016;32(5):435–40.

    Article  PubMed  Google Scholar 

  15. Afshar M, Lanoue A, Sallantin J. Multiobjective/multicriteria optimization and decision support in drug discovery. Comprehensive Medicinal Chemistry II2007. p. 767–74. https://doi.org/10.1016/b0-08-045044-x/00275-3.

  16. Breuer R, Mattheisen M, Frank J, Krumm B, Treutlein J, Kassem L, et al. Detecting significant genotype–phenotype association rules in bipolar disorder: market research meets complex genetics. Int J Bipolar Disord. 2018;6(1):24.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Williams C, Polom K, Adamczyk B, Afshar M, D’Ignazio A, Kamali-Moghaddam M, et al. Machine learning methodology applied to characterize subgroups of gastric cancer patients using an integrated large biomarker dataset. EJSO. 2019;45(2):e75.

    Article  Google Scholar 

  18. Hampel H, Williams C, Etcheto A, Goodsaid F, Parmentier F, Sallantin J, et al. A precision medicine framework using artificial intelligence for the identification and confirmation of genomic biomarkers of response to an Alzheimer’s disease therapy: analysis of the blarcamesine (ANAVEX2-73) Phase 2a clinical study. Alzheimers Dement (N Y). 2020;6(1): e12013.

    Article  PubMed  Google Scholar 

  19. Liquiere M, Sallantin J. Structural machine learning with galois lattice and graphs. Proc Fifteenth Int Conf Mach Learn. 1998;98:305–13.

    Google Scholar 

  20. Gasmi G, Yahia SB, Nguifo EM, Slimani Y, editors. A new informative generic base of association rules. 2005; Berlin, Heidelberg: Springer Berlin Heidelberg. https://doi.org/10.1007/11430919_11.

  21. Nebot V, Berlanga R. Finding association rules in semantic web data. KBS. 2012;25(1):51–62.

    Google Scholar 

  22. Zhao Y, Zhang H, Figueiredo F, Cao L, Zhang C. Mining for combined association rules on multiple datasets. Proceedings of the 2007 international workshop on Domain driven data mining - DDDM '072007. p. 18–23. https://doi.org/10.1145/1288552.1288555.

  23. Kuznetsov SO, editor. Machine learning and formal concept analysis. 2004; Berlin, Heidelberg: Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-540-24651-0_25.

  24. Payne PRO, Mendonça EA, Johnson SB, Starren JB. Conceptual knowledge acquisition in biomedicine: a methodological review. J Biomed Inform. 2007;40(5):582–602.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Indrayan A, Mishra A. The importance of small samples in medical research. J postgrad med. 2021;67(4):219–23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Greenbaum CJ, Beam CA, Boulware D, Gitelman SE, Gottlieb PA, Herold KC, et al. Fall in C-peptide during first 2 years from diagnosis: evidence of at least two distinct phases from composite Type 1 Diabetes TrialNet data. Diabetes. 2012;61(8):2066–73.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Durinovic-Belló I, Gersuk VH, Ni C, Wu R, Thorpe J, Jospe N, et al. Avidity-dependent programming of autoreactive T cells in T1D. PLoS ONE. 2014;9(5):e98074.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Yang J, Danke N, Roti M, Huston L, Greenbaum C, Pihoker C, et al. CD4+ T cells from type 1 diabetic and healthy subjects exhibit different thresholds of activation to a naturally processed proinsulin epitope. J Autoimmun. 2008;31(1):30–41.

    Article  PubMed  Google Scholar 

  29. Alhadj Ali M, Liu YF, Arif S, Tatovic D, Shariff H, Gibson VB, et al. Metabolic and immune effects of immunotherapy with proinsulin peptide in human new-onset type 1 diabetes. Sci Transl Med. 2017;9(402):eaaf777.

    Article  Google Scholar 

  30. Grissa D, Comte B, Pétéra M, Pujos-Guillot E, Napoli A. A hybrid and exploratory approach to knowledge discovery in metabolomic data. Discrete Appl Math. 2020;273:103–16.

    Article  Google Scholar 

  31. Bundy BN, Krischer JP. A quantitative measure of treatment response in recent-onset type 1 diabetes. Endocrinol diabetes metab J. 2020;3(3): e00143.

    Article  CAS  Google Scholar 

  32. Dennis JM, Shields BM, Henley WE, Jones AG, Hattersley AT. Disease progression and treatment response in data-driven subgroups of type 2 diabetes compared with models based on simple clinical features: an analysis using clinical trial data. Lancet Diabetes Endocrinol. 2019;7(6):442–51.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Herold KC, Bundy BN, Long SA, Bluestone JA, DiMeglio LA, Dufort MJ, et al. An anti-CD3 antibody, teplizumab, in relatives at risk for type 1 diabetes. NEJM. 2019;381(7):603–13.

    Article  CAS  PubMed  Google Scholar 

  34. Edner NM, Heuts F, Thomas N, Wang CJ, Petersone L, Kenefeck R, et al. Follicular helper T cell profiles predict response to costimulation blockade in type 1 diabetes. Nat Immunol. 2020;21(10):1244–55.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Hannelius U, Beam CA, Ludvigsson J. Efficacy of GAD-alum immunotherapy associated with HLA-DR3-DQ2 in recently diagnosed type 1 diabetes. Diabetologia. 2020;63(10):2177–81.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Hahsler M, Grün B, Hornik K. arules - a computational environment for mining association rules and frequent item sets. J Stat Softw. 2005;14(15):1–25.

    Article  Google Scholar 

  37. Gianani R, Eisenbarth GS. The stages of type 1A diabetes: 2005. Immunol Rev. 2005;204:232–49.

    Article  CAS  PubMed  Google Scholar 

  38. Battaglia M, Ahmed S, Anderson MS, Atkinson MA, Becker D, Bingley PJ, et al. Introducing the endotype concept to address the challenge of disease heterogeneity in type 1 diabetes. Diabetes Care. 2020;43(1):5–12.

    Article  PubMed  Google Scholar 

  39. Ludvigsson J, Samuelsson U, Beauforts C, Deschamps I, Dorchy H, Drash A, et al. HLA-DR 3 is associated with a more slowly progressive form of type 1 (insulin-dependent) diabetes. Diabetologia. 1986;29(4):207–10.

    Article  CAS  PubMed  Google Scholar 

  40. Hall FC, Rabinowitz JD, Busch R, Visconti KC, Belmares M, Patil NS, et al. Relationship between kinetic stability and immunogenicity of HLA-DR4/peptide complexes. Eur j immunol. 2002;32(3):662–70.

    Article  CAS  PubMed  Google Scholar 

  41. Bediaga NG, Garnham AL, Naselli G, Bandala-Sanchez E, Stone NL, Cobb J, et al. Cytotoxicity-related gene expression and chromatin accessibility define a subset of CD4+ T cells that mark progression to type 1 diabetes. Diabetes. 2022;71(3):566–77.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Sano H, Imagawa A. Pancreatic beta-cells express major histocompatibility complex class II: Do diabetic beta-cells have the capacity of antigen-presenting cells? J Diabetes Investig. 2020;11(2):281–3.

    Article  PubMed  Google Scholar 

  43. Arif S, Tree TI, Astill TP, Tremble JM, Bishop AJ, Dayan CM, et al. Autoreactive T cell responses show proinflammatory polarization in diabetes but a regulatory phenotype in health. J Clin Invest. 2004;113(3):451–63.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Cerboni S, Gehrmann U, Preite S, Mitra S. Cytokine-regulated Th17 plasticity in human health and diseases. Immunology. 2021;163(1):3–18.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

The authors would like to thank the patients and their families, all participating investigators, sites, and site teams. Kezia Pittaway of inScience Communications, Springer Healthcare, UK, provided medical writing support, which was funded by Imcyse in accordance with Good Publication Practice 2022 guidelines.

Funding

Imcyse S.A. (Liège, Belgium), EU FP7 program (N°602515), DG06 (Walloon Region, Belgium, N°7873).

Author information

Author notes

  1. Jean Van Rampelbergh, Peter Achenbach, Richard David Leslie, Pierre Vandepapelière and Christian Boitard are senior authors.

Authors and Affiliations

  1. Imcyse S.A, Avenue Pré-Aily 14, Liège, 4031, Belgium

    Jean Van Rampelbergh, Vincent Carlier, Nicolas Bovy, Luc Vanderelst, Marcelle Van Mechelen & Pierre Vandepapelière

  2. Institute of Diabetes Research, Helmholtz Zentrum München, German Research Center for Environmental Health, Munich-Neuherberg, Germany

    Peter Achenbach

  3. Forschergruppe Diabetes, Technical University Munich, Klinikum Rechts Der Isar, Munich, Germany

    Peter Achenbach

  4. Department of Immunobiology, Queen Mary University of London, London, UK

    Richard David Leslie

  5. Ariana Pharmaceuticals SA, Paris, France

    Martin Kindermans & Frédéric Parmentier

  6. Inserm U1016, Cochin Institute, Paris, France

    Christian Boitard

  7. Medical Faculty, Université de Paris, Paris, France

    Christian Boitard

Authors
  1. Jean Van Rampelbergh
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Peter Achenbach
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Richard David Leslie
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Martin Kindermans
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Frédéric Parmentier
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Vincent Carlier
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Nicolas Bovy
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Luc Vanderelst
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Marcelle Van Mechelen
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Pierre Vandepapelière
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Christian Boitard
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

JVR, VC, NB, LV, MVM, PV: laboratory analysis; MK, FP: machine learning analysis; JVR, PA, RDL, NB, CB: clinical study concept and design; PA, RDL, CB: patient recruitment and treatment. All authors: data analysis and interpretation, critical review of the manuscript, approval of the final version of the manuscript.

Corresponding author

Correspondence to Jean Van Rampelbergh.

Ethics declarations

Ethics approval and consent to participate

The study was conducted across multiple countries and institutions and was approved by all relevant ethics committees and national authorities. Belgium: Medical Ethics Committee UZ Brussel—Ref2017/122, Denmark: Capital Region of Denmark—Centre for Health—Scientific ethics Committee—Ref: H-17009698, France: Comité de Protection des Personnes Sud Méditerranée I—Ref 17 45, Germany: Technical University of Munich—Faculty of Medicine—Ethics Committee—Ref 128/17 Af, Lithuania: Lithuanian Bioethics Committee—Ref 2018–03–21 No. P-I 8–20/2, Sweden: Regional Ethical Review Board in Gothenburg—Ref 189–18, United Kingdom: London—Chelsea Research Ethics Committee—Ref 17/LO/0633. All patients signed informed consent before taking part in the study.

Consent for publication

The data of individual patients were de-identified and used for statistical analysis; no individual patient data are presented in the manuscript.

Competing interests

JVR, VC, NB, LV, MVM, and PV are employees or contractors of Imcyse S.A., Liège, Belgium, and may hold stock options. MK and FP are employees of Ariana Pharmaceuticals SA. RDL’s institution received study funding and materials from Imcyse. RDL received honorarium from DMRR and took part in advisory boards for Diamyd and Provention. CB’s institutions received study funding and materials from Imcyse. PA declares no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

12916_2024_3476_MOESM1_ESM.docx

Additional file 1: Supplementary Material. Fig. S1. Overview of main steps used to identify most relevant relationships between variables. Fig. S2. Immune response to IMCY-0098 treatment. Fig S3. Associations between clinical and immune response parameters. Fig. S4. Evolution of CD4+/Granzyme B+ CD107a cells in IMCY-0098 treated participants compared to placebo. Table S1. HLA haplotypes by cohort. Table S2. Formal concept analysis results. Table S3. Markers used for detection of cytolytic and pathogenic T cells combined with multiparameter flow cytometry.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Van Rampelbergh, J., Achenbach, P., Leslie, R.D. et al. First-in-human, double-blind, randomized phase 1b study of peptide immunotherapy IMCY-0098 in new-onset type 1 diabetes: an exploratory analysis of immune biomarkers. BMC Med 22, 259 (2024). https://doi.org/10.1186/s12916-024-03476-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12916-024-03476-y

Keywords

BMC Medicine

ISSN: 1741-7015

Contact us